slick agent¶

Agent tooling. agent schema emits the full command tree as JSON; agent guide emits Markdown runbooks scoped to specific workflows. Both are designed to be consumed by LLM agents rather than humans.

slick agent schema  Output command schema as JSON
slick agent guide   Output workflow instructions for agents

agent schema¶

Emit the command tree, flag inventory, output modes, and exit-code contract as JSON.

slick agent schema                          # full schema, full envelope
slick agent schema --compact                # minimal schema (smaller records)
slick agent schema --output=compact         # full schema, no envelope (data only)
slick agent schema --compact -o compact     # minimal schema, no envelope

Flags¶

-c, --compact  Emit a minimal schema (smaller per-command records)

--compact here is the command-local schema-shape flag (short form -c). It is distinct from the global --output=compact (short form -o compact), which strips the JSON envelope. Both can be combined; use -o compact to toggle the envelope independently.

The schema's top-level object includes:

version, description — schema version and a human-readable summary.
auth — supported auth shapes and required scopes.
output — output modes (--output=auto|human|json|compact).
global_flags — flags accepted on every command (workspace selection, output mode, debug, color, timeout, throttle). Attribution toggles (--attribution, --no-attribution, --attribution-{label,emoji,message}) are command-local on the four mutating commands (message send, message edit, reply, file upload) and appear under each command's flags entry rather than global_flags.
commands — the full command tree, with each entry carrying name, full_path, description, read_only, optional flags, and a recursive subcommands array.
input_shapes, output_schemas — JSON-shape descriptors for the structured inputs (e.g. Block Kit) and outputs each command emits.
env — environment variables slick reads, including config overrides and the agent-detection trigger set.
exit_codes — the 1..7 mapping with errors[0].type names.
examples — annotated invocation examples per command.
workflows — pointers into the agent guide workflow catalogue.
best_practices, anti_patterns — guidance an agent should follow or avoid (e.g. "Slack timestamps are channel-scoped strings — keep them as strings", "do not synthesise channel IDs from DM contexts").

This file is the source of truth for agents that need to build a command plan without prompting the user.

agent guide¶

Emit Markdown runbooks for specific workflows. Each runbook is a short preconditions / commands / parse / quirks sequence.

slick agent guide --help          # list available workflows
slick agent guide send_msg        # one specific workflow
slick agent guide schedule_msg    # scheduled send/list/delete workflow
slick agent guide                 # the full guide (every workflow)

Available workflows¶

auth_setup            Generate a manifest and authenticate a profile
cache_metadata        Prime users and channels for repeated lookup and shell completion
check_health          Check Slack service health and Web API reachability
cleanup_msgs          Runbook for cleaning test messages found through paginated search
config_prefs          Set profile preferences without touching auth
core_contract         Understand output modes, stderr/stdout, and fixed exit codes
delete_msg            Delete own messages with dry-run and force safeguards
developer_review      Runbook for posting a review-style message with reactions and a thread
discover_destination  Find channel and DM IDs before posting
edit_msg              Edit own messages by exact channel and timestamp
inspect_schema        Read the machine schema and workflow guide
lookup_user           Find user IDs, presence, status, and timezone
react                 Add, remove, and list emoji reactions by channel and timestamp
read_history          Read channel history or thread replies with bounded pagination
reply                 Reply to a message thread by parent timestamp
safe_mutation         Preview high-impact changes and parse JSON results
schedule_msg          Schedule, list, and delete future messages
search_msgs           Workspace message search workflow
send_dm               Send direct messages while handling token limits
send_msg              Send a markdown message and read ts/permalink from JSON
set_status            Set or clear the authenticated user's Slack status
upload_file           Probationary hidden file upload workflow

Tone¶

Each runbook is direct, command-shaped, and stateless. The intent is machine-parseable instructions an agent can follow without having to re-read the whole project. The runbooks track the same JSON shapes documented in the per-command pages of this site, including post-v0.4.0 renames (ts, id/name, dropped action-result bools, ClearData split).